Infants with severe CDH are intubated immediately after birth [10, 11], but the optimal initial mode of ventilation remains unclear for this population. In this study, we leveraged our high-volume population of infants with severe CDH to address this evidence gap. In our experience, HFOV as the initial mode of ventilation in the DR for infants with severe CDH was associated with improved gas exchange during the postnatal transition without observed differences in adverse hospital outcomes.

The existing literature for initial mode of ventilation for infants with CDH is inconsistent. The VICI trial randomized 91 infants with CDH to CMV compared to 80 infants to HFOV. While there was no significant difference in the combined outcome of mortality or chronic lung disease, infants randomized to HFOV experienced higher ECMO use and longer duration of mechanical ventilation [8, 9]. Conversely, Fuyuki et al. conducted a multicenter retrospective cohort study of initial mode of ventilation within the first 24 h for 135 infants with CDH. Following propensity score matching of 45 infants treated with CMV and 90 with HFOV, there were no differences in outcomes of ECMO, mortality, or chronic lung disease [12]. Finally, Semama et al. compared outcomes of infants with CDH between 2 French centers; one center prioritized CMV and the other prioritized HFOV using low MAP < 12 cm H2O. Duration of oxygen therapy and survival did not differ between centers, but infants treated in the center that prioritized HFOV experienced longer duration of mechanical ventilation and sedation. Of note, sedation and hemodynamic management differed between the two centers [13].

In the present study, initial mode of ventilation was not associated with mortality, ECMO, or duration of ventilation after adjusting for CDH side and O/E LHR. A notable difference between the VICI trial and ours was CDH severity. In our study, the median O/E LHR was 26.5% in the HFOV group and 36.6% in the CMV group compared to 47% in the HFOV group and 48% in the CMV group in the VICI trial. This difference is important as mortality and ECMO utilization are known to be higher in infants with more severe CDH [7, 14, 15]. Despite the high severity in our cohort, overall mortality was similar to the CMV group in the VICI trial who had less severe CDH. Our more contemporary HFOV group also had more severe CDH. To account for this, models were adjusted for CDH side and O/E LHR.

Ventilator settings are another important component to consider, particularly the mean airway pressure and the balance between achieving lung inflation and avoiding overdistension once aerated. In the VICI trial, initial CMV settings were PIPs of 20–25 cm H2O and PEEP of 3–5 cm H2O, with a ventilator rate of 40–60/min. These are very similar to our CMV settings with the exception of PEEP, which was consistently set at 5–6 cm H2O at our center. In contrast, the HFOV settings used in our DR differed substantially from those used in the VICI trial. In the VICI trial, initial HFOV settings included a MAP 13–17 cm H2O, frequency of 10–12 Hz, and amplitude of 30–50 cm H2O depending on chest wall vibration. The MAP settings used for HFOV in our study were lower than those used in VICI; the median MAP was 13 cm H2O, frequency was 7 Hz, and amplitude was 33 cm H2O. We speculate that HFOV settings used in our cohort may be more lung protective than those used in the VICI trial. This is important as MAP settings below 13 cm H2O have been associated with lower ECMO, vasoactive drugs, and inhaled nitric oxide use [16].

Previous studies have demonstrated hypercarbia after birth to be associated with increased mortality and ECMO use [2,3,4,5,6,7]. In our study, HFOV was associated with improved gas exchange immediately after birth. Although this did not translate to benefit for ECMO or mortality, preventing severe acidosis may have a beneficial impact on cerebral hemodynamics, pulmonary hypertension, cardiac function, and neurodevelopment. Future areas of study may assess the impact of severe hypercarbia and acidosis on neurologic and cardiac outcomes among infants with CDH.

Our study has limitations and unique strengths. This was an observational cohort study and therefore a non-randomized sample. We adjusted for known confounders, such as CDH side and severity in our analysis. We chose to evaluate hospital based ventilatory outcomes as those were the outcomes reported in the VICI trial. We acknowledge that other important outcomes such as cardiac function, pulmonary hypertension, and other complications were not evaluated. In Supplementary Table 2, we compare vasoactive support and pulmonary hypertension treatment between the two groups. While there were no differences in the use of inhaled nitric oxide, dopamine infusions, or use of pulmonary hypertension medications at discharge, there was increased epinephrine and alprostadil use in the more contemporary HFOV cohort. Use of epinephrine infusions was intentionally increased in an effort to decrease use of high dose dopamine infusions. These differences reflect practice changes in our center that took place following DR resuscitation. Echo-guided hemodynamic management was not a standardized practice in our center across both epochs. We also did not report long term neurodevelopmental outcomes. In addition, we defined our cohort to ensure there were no other important changes in DR clinical practice during the study period. However, it is possible that unmeasured confounders differed between groups. As a single center study, outcomes may also reflect practice management differences unique to our center. Finally, we used a sample of convenience based on the number of infants treated with each ventilator during the study period; our sample size was similar to that of the VICI trial. It is possible that we were underpowered to detect differences in clinical outcomes between groups. Study strengths include one of the largest cohorts of infants with CDH comparing CMV and HFOV. In addition, we care for a large population of infants with severe CDH, and this study fills an important evidence gap specific to this high-risk population that is not always represented in clinical trials.